Making good predictions

In his book What the dog saw: and other adventures, prolific writer and thinker Malcolm Gladwell devotes an entire part to theories, predictions and diagnoses. Predicting something accurately isn’t always easy. Gladwell explains this by touching on the difference between puzzles and mysteries. This idea originates from Gregory Treverton, an authority in the field of intelligence, terrorism and US national security. Puzzles are questions with answers. Usually we cannot solve them straightaway because we have too little information available, or we miss the most crucial information. In order to solve the puzzle, we must gather more information. Mysteries, in contrary, deal with situations where we have an abundance of information available. No one really has the answer. If we want to solve mysteries, we need to come up with judgements and estimates. It therefore requires a different approach and skill set to solve these kind of problems.

Treverton suggests that we need to think of puzzles as facts and mysteries as intentions, so that we can see how one could design systems to discover near-infinite amounts of facts while still being unable to discern a target’s intentions from those facts. I believe that we always need to keep this difference in mind. In fact, it is a good framework to think in the area of  Service Management where I am currently working in. When a client files a ticket, in some cases it might be very hard to make a prediction in terms of incident diagnosis, how fast it can be resolved, setting the right expectations et cetera. Even detecting or replicating an issue might take more time than initially thought of. Sometimes you are sifting through a heap of information (i.e., a mystery) and as a result you are looking in the complete wrong direction.

Who is needed is someone who is able to make good (read: not correct)judgements, be it by experience or through applying heuristics or theory. For instance,  sometimes it is better to classify a scenario immediately as a problem in order not to miss the Service Level Agreement and not waste team in trying to understand the issue in more detail and trying to understand the root cause.

Another point is that you need to be able to ask the right questions and prioritize in terms of urgency. People on the ground or in the field, closest to the technology, are to gather the evidence, but it is equally important to verify and review information accurateness. What I think is key in this story is that information should be able to spread and connect freely throughout the intelligence gathering. However, in order to make good estimates and judgements, ask the right questions and so fort, a system should be designed with the outcome in mind. For instance, if you already think of how many customers, suppliers, orders and parts there can possibly exist, you can make better decisions upfront of how to display them in your information systems. You also need to be able to classify things in mutually exclusive and collectively exhaustive categories and use these as your map.

To end with, I believe a good policy maker is one who understands the global picture and see how if fits with other processes. He has to see through the intelligence process, has to be well-informed, vastly experienced, and constantly be thoughtful enough to make things better.

Leave a Reply

Your email address will not be published. Required fields are marked *